What Everyone Needs to Know About Computation
John Barr Steve Cooper Mike Goldweber
Ithaca College Purdue University
barr@ithaca.edu coopers@acm.org
mikeyg@cs.xu.edu
Categories and Subject Descriptors
K.3.2 [Computers and Education]: Computer and Infor-
mation Science Education—Computer Science Education
General Terms
Design, Experimentation
Keywords
Non-major courses, Curriculum, Computational Thinking
1. SUMMARY
For a variety of reasons there has been a renewed interest
in non-major level computer science courses. One reason
for this is that non-major courses are seen as potentially in-
creasing both interest and enrollment in computer science
courses. Another reason is the rising popularity of “compu-
tational thinking.” Furthermore, as computation becomes
the cornerstone of research methodologies, if not professional
practice, in the natural, physical and social sciences, includ-
ing, as well, a growing influence in the humanities, the de-
mand from external departments/programs for non-major
level service courses has began to rise, albeit slowly.
This raises some interesting questions to the CSE commu-
nity. Given that one offers a single non-major level course,
what should be the topics covered as distinct from the tech-
nologies used to convey those topics? What learning out-
comes are most desired? Finally, the competency, or at
least serious introduction of which technological skills, if any,
should be considered mandatory (e.g. programming).
In short, what should a non-major/minor student who
takes one course in computer science come away with know-
ing and being able to do? Possibly orthogonal to this is
the question: What should any liberally (college) educated
person know about computation at this point in time?
This panel will outline four different perspectives on this
important question in addition to engaging the audience in
a discussion regarding the answers to the questions raised.
2. JOHN BARR
Our experience at Ithaca College is that students entering
from high school have, over the last 10 years, become less
Copyright is held by the author/owner(s).
SIGCSE’10, March 10–13, 2010, Milwaukee, Wisconsin, USA.
ACM 1-58113-798-2/04/0003.
Henry Walker
(Moderator) Grinnell College
Xavier University walker@math.grinnell.edu
computer literate. Ten years ago we could assume that stu-
dents knew what a folder or directory was, how to create,
rename, and move files, how to install programs, and even
how to change basic operating system settings. Today we
find that none of these skills are prevalent. Instead, students
are experts at finding papers on the Web, installing apps on
their cell phones, and are masters at all sorts of communi-
cation and social networking technologies from cell phones
to texting to Facebook.
The traditional view of computer literacy sees this trend
as an unfortunate development that needs to be reversed
through computer literacy/non-major level courses. Stu-
dents should be forced to learn computation, abstraction
(such as folders and files), and the basics of maintaining
their operating system. But is this true? What do students
really need to know?
The discipline of computer science is dedicated to develop-
ing computational tools. One aspect of this goal is to make
the tools more accessible and flexible. As the discipline ma-
tures, the tools it produces have also become more mature.
Compare, for example, the iPhone against all previous smart
phones. Certainly the iPhone is no less useful or powerful
than previous phones. Yet users no longer need to know the
arcania of folders, the simultaneous use of multiple buttons,
and stupefying navigation techniques.
The lesson to take from this simple look at cell phone tech-
nology is not that users are becoming less adept. Rather the
lesson is that users have mastered the technology that exists
today, not the technology that existed ten years ago. The
gritty details of computation and communication have been
abstracted out and users no longer need to master these de-
tails to use these tools effectively. Just as some people would
argue that computer scientists do not need to be masters of
assembly language programming to be successful today, the
average user of technology no longer needs to master the de-
tails of computation to be an effective user of computation-
based technology.
3. MICHAEL GOLDWEBER
It is instructive to consider a proposed course, not just
against other proposed courses, but against no course at all.
After all, both automotive and telephone technology are as
ubiquitous as computation, yet the vast majority can suc-
cessfully utilize both technologies without any grounding,
i.e. non-majors college course, in either physics or mechan-
ical engineering. Taking this argument further one might
conclude that other disciplines which require knowledge of
computation would be satisfied by either requiring an intro-
127
ductory major-level CS course or covering what aspects of
computation are needed in their own courses.
This position argues that a meaningful non-major level
CS course should therefore not be concerned with skill ac-
quisition. Instead it should focus on the discipline’s deep
intellectual contributions; which may be described as algo-
rithmic problem solving, harnessing abstraction to manage
complexity, and the automation of an algorithmic solution.
Of these, I believe that algorithmic problem solving is not
only the most important, but also the one where the greatest
success can be achieved.
Algorithmic problem solving has a number of standard
paradigms; brute force, greedy, divide and conquer, dynamic
programming, etc. In addition to these techniques are the
important notions of problem reduction, when a new prob-
lem can be recast as an instance of a different, previously
solved problem, and the recognition and handling of “hard”
problems. Whether or not one ever intends to automate a
given problem’s solution, being able to approach and even-
tually solve a given problem using one or more of these tech-
niques is a highly valued skill.
Consider a non-major level course that involves no pro-
gramming, no discussion of objects, encapsulation or poly-
morphism, but examines a wide variety of problems drawn
from a plethora of problem domains and has students prac-
ticing how to develop solutions to these problems. Just as
a liberally educated person should be versed in how mathe-
maticians determine truth/solves problems; conjecture and
proof, or how natural and social scientists do the same; re-
producible experiments informed via the scientific method,
they should also be versed in how computer scientists do
the same; application of problem reduction and/or the use
of standard algorithmic paradigms.
4. HENRY WALKER
Much discussion by panelists here and by others through-
out the computing community supports the view that intro-
ductory computing courses should involve problem solving.
In principle, I too support the idea that problem solving
should be central to computer science courses — for non-
majors, majors, and anyone else. In courses, I would hope
that students would tackle problems and gain experience at
finding solutions. However, I wonder how students (or the
general citizenry) will state and explain their solutions, and
how they will know their proposed solutions are correct.
I think much is lost if people (students, the general public)
do not have to be precise in stating their solutions. It is rea-
sonably easy to write high-level outlines for many problems
in English, but it does not follow that these approaches can
be refined to actually work in practice. Algorithmic problem
solving illustrates the maxim, “the devil is in the details.”
This difficulty leads to a dilemma. Although courses for
both majors and non-majors should emphasize problem solv-
ing, omitting some type of programming allows people to be
sloppy. This observation leads to the conclusion that pro-
gramming of some type is needed to sharpen problem-solving
skills.
Of course, one can become caught up in details of pro-
gramming languages (some courses seem overly focused here),
but there needs to be some way to make solutions precise.
Whatever the language, students should be able to write
careful and detailed solutions, and the language should have
sufficient expressibility and precision that students and their
128
instructors can analyze the correctness of any proposed solu-
tions. While not essential, machine-executable instructions
also can be helpful in determining reviewing correctness.
In this regard, pseudo-code is no better than a formally-
defined programming language. If the pseudo-code is defined
tightly enough, then it might as well be machine readable.
If pseudo-code is not tightly defined, then it is open to the
same ambiguities as English or another natural language.
Overall, I conclude introductory courses for both non-majors
and majors should include a significant component of pro-
gramming as a mechanism to express ideas carefully and
precisely.
5. STEVE COOPER
J. Wing has indicated that she considers computational
thinking to be the process of abstraction. Though Wing goes
on to focus on the “goodness” of abstractions from an engi-
neering point of view (examining such measures as efficiency,
correctness, and software engineering considerations such as
simplicity, elegance, usability, modifiability, maintainability,
cost, etc.), the bottom line is that any approach towards pre-
senting computational thinking for the masses needs to focus
on the process of abstraction. Of course, teaching the pro-
cess of abstraction is a daunting task, attempting to tackle
one of Brooks’ essential difficulties (referred to as software
invisibility) inherent in software engineering.
If a non-majors course needs to help students understand
the process of abstraction, it seems to me to be necessary to
provide the students meaningful contexts from which they
are comfortable in exploring abstraction. Certainly my own
work with Alice (www.alice.org, www.aliceprogramming.
net) provides the context of animation and storytelling from
which to explore abstraction. Scratch (scratch.mit.edu) is
an alternative. Guzdial’s work with media computation uses
the manipulation of sound, images, and video as an alter-
native context from which to explore abstraction. Roberts’
work with the ACM Java Taskforce provides instructors with
a powerful library of graphics and other routines from which
to construct their own contexts. Robots are another con-
text, as is the programmable clothing of Lilypad Arduino.
And there are many others. It is also not necessary to fo-
cus on programming in providing the context. It is cer-
tainly possible to explore abstraction through the contexts
of networking and/or communication. It is also possible to
explore computational thinking and the process of abstrac-
tion through domain specific examples/contexts from the
geo-sciences, economics, politics, biology, physics, etc. In-
structors can demonstrate how computational thinking can
be used to provide abstractions for the modeling of behavior.
Guzdial’s media computation is used as an example con-
text by which to explain what I mean by the exploration of
abstraction. A picture is an abstraction for a collection of
pixels. A pixel is an abstraction for a 3-tuple consisting of
x and y coordinates and a color. A color is an abstraction
for a 3-tuple, say of red, green, and blue. And the abstrac-
tions can continue, by considering how a collection of images
can be considered as a video, etc. The bottom line is that
a concrete example, meaningful to the student, provides a
potential context from which to help students explore the
process of abstraction.